E demli, Candan; Vall Cas elló, Judi
A icle
Gende di e ences in online educa ion
SERIEs - Jou nal o he Spanish Economic Associa ion
P o ided in Coope a ion wi h:
Spanish Economic Associa ion
Sugges ed Ci a ion: E demli, Candan; Vall Cas elló, Judi (2024) : Gende di e ences in online
educa ion, SERIEs - Jou nal o he Spanish Economic Associa ion, ISSN 1869-4195, Sp inge ,
Heidelbe g, Vol. 15, Iss. 4, pp. 349-388,
h ps://doi.o g/10.1007/s13209-024-00299-1
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/326960
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
SERIEs (2024) 15:349–388
h ps://doi.o g/10.1007/s13209-024-00299-1
ORIGINAL ARTICLE
Gende di e ences in online educa ion
Candan E demli1
·Judi Vall Cas elló2
Recei ed: 10 July 2023 / Accep ed: 25 June 2024 / Published online: 5 Augus 2024
© The Au ho (s) 2024
Abs ac
Online lea ning esou ces ha e become ex emely popula , pa icula ly a e he
es ic ions caused by he Co id-19 ou b eak. In his pape , we use da a o Spain
a he indi idual le el om an online ma h lea ning pla o m which is used by chil-
d en om o e 100 coun ies, o documen he gende di e ences in he con ex o
online lea ning. We quan i y he gende gaps in e o and ela i e pe o mance ou -
comes and analyze whe he he gaps di e by he gende o he pa en who mainly
supe ises he child en. Ou main esul s poin owa d signi ican gende gaps in he
ela i e pe o mance ou comes in a o o boys, while he e idence o he e o
gende gaps is only signi ican when we compa e he siblings o he opposi e gende
(con olling o pa en ixed e ec s). Fu he , we ind ha li ing in municipali ies wi h
mo e egali a ian gende no ms is associa ed wi h na owe o posi i e gende gaps
in e o ou comes, while we do no ind such di e ences in he ela i e pe o mance
ou comes. Taking in o accoun he inc ease in he use o online lea ning ools and
hei p og essi e in eg a ion in o he egula educa ional sys em, ou esul s p o ide
impo an in o ma ion o minimize gende biases in hese new se ings.
Keywo ds Educa ion ·Gende ·Online lea ning
We a e g a e ul o pa icipan s a he IWIP Semina s a IEB, Uni e si y o Ba celona; he 34 h Annual
Con e ence o he Eu opean Socie y o Popula ion Economics (ESPE); AEET - Labo Economics
Mee ing 2021; and he Applied Economics Con e ence: Labou , Heal h, Educa ion, and Wel a e; and o
Libe ad González o hei help ul commen s. We would like o exp ess ou app ecia ion o Sma ick
Company o gene ously p o iding us wi h he da ase used in his s udy. We would like o acknowledge
inancial suppo om Ho izon Eu ope - Eu opean Resea ch Execu i e Agency (REA)
HORIZON-CL2-2022-DEMOCRACY-01; p ojec numbe 101095106: “Educa ion o Responsible
Democ a ic Ci izenship (DEMOCRAT).” Candan E demli acknowledges inancial suppo om he
FI-AGAUR. The eplica ion ma e ial o he s udy is a ailable a 10.5281/zenodo.11314201.
BCandan E demli
[email p o ec ed]
Judi Vall Cas elló
judi . [email protected]
1Depa men o Economics and Business, Cen al Eu opean Uni e si y and IEB, Quellens aße
51, Wien 1100, Aus ia
2Depa men o Economics, Uni e si y o Ba celona, IEB and CRES-UPF, J.M. Keynes, 1, 11, 08034
Ba celona, Spain
123
350 SERIEs (2024) 15:349–388
JEL Classi ica ion J13 ·J16 ·I24
1 In oduc ion
Du ing he las ew yea s, he use o online lea ning ools in educa ion has been
on he ise in mos de eloped coun ies. This end was u he accele a ed by he
Co id-19 ou b eak and he subsequen school closu es in 2020. Gi en he inc easing
p e alence o hese ools and hei likely in eg a ion in o he mains eam educa ion
sys em, i is essen ial o unde s and he ole played by pa en s and he e o s pu
o h by child en. On he o he hand, p e ious economics li e a u e has documen ed
gende di e ences in pa en al in es men , no only in de eloping coun ies bu also in
he con ex o de eloped coun ies. In he USA, boys ecei e mo e pa e nal ime han
gi ls (Lundbe g e al. 2007;P ice2008). In Canada, he UK, and he USA, pa en al ime
in es men s in eaching ac i i ies, such as eading, end o a o gi ls, while a he s
in es signi ican ly mo e ime in boys (Bake and Milligan 2016). Di e en ial pa en al
in es men o boys and gi ls migh a y be ween mo he s and a he s as well. Mammen
(2011) inds ha in he USA, a he s alloca e mo e ime o hei child en i hey ha e
a leas one boy, whe eas mo he s’ o al ime in es men is he same ega dless o he
gende composi ion o hei child en. Fu he mo e, no only he ime in es ed may be
di e en acco ding o he gende ma ch be ween pa en s and hei child en bu also
o he elemen s a ec ing he educa ional ou comes may be di e en depending on his
gende ma ch. I simila pa en al gende bias and di e ences a e p esen when using
online lea ning ools, i may lead o a u u e gende gap in educa ional achie emen
and labo ma ke ou comes.
This pape analyzes he gende gaps in online lea ning and how hese gaps a e
co ela ed wi h he gende o he main supe iso . In pa icula , we quan i y he e o
and pe o mance gaps be ween gi ls and boys when using an online lea ning pla o m
in Spain and desc ip i ely analyze whe he he di ec ion and he magni ude o hese
gaps di e depending on he gende o he pa en who mainly supe ises he child.
Fu he mo e, we analyze whe he he esul s a e he e ogeneous by he gende o he
eldes sibling using he pla o m ( o es o any ole model e ec ), and he gende
no ms in he municipali y o esidence.
We ind e idence o signi ican gende gaps in he ela i e pe o mance ou comes
in a o o boys, while he e idence o he e o gende gaps is only signi ican when
we compa e he siblings o he opposi e gende . We also ind sugges i e e idence ha
he e o gaps a e na owe —e en posi i e in a o o gi ls o pe sis ency ou come
which indica es whe he a leas one session is comple ed in each mon h—when he
main supe iso is he mo he . Fo pe o mance ou comes, on he o he hand, we
do no obse e any di e ences in he gende gap depending on he gende o he
main supe iso . We conduc he same analysis wi hin a subsample o siblings o he
opposi e gende , including pa en ixed e ec s. While gende gaps in pe o mance
ou comes in sign and magni ude a e e y simila o hose in he ull sample, we do
no obse e any signi ican di e ences in he gende gaps (nei he o e o no o
pe o mance ou comes) depending on he gende o he main supe iso in he siblings’
subsample. We ind he e ogenei y by he gende o he eldes sibling in bo h a he s’
123
SERIEs (2024) 15:349–388 351
and mo he s’ samples. Speci ically, while he gende gap in he ex a ime de o ed o
sol ing p oblems and he numbe o sessions comple ed pe mon h come om amilies
whe e he eldes is a gi l, he gende gap in he a io o co ec ly sol ed p oblems is
mo e p onounced in amilies whe e he eldes is a boy. Addi ionally, we ind ha li ing
in a municipali y wi h mo e egali a ian gende no ms is associa ed wi h gende gaps
in a o o gi ls in e o ou comes. Howe e , we do no obse e such a di e ence in
gende gaps in ela i e pe o mance ou comes. The gende gap in delayed comple ion
which ep esen s he ex a ime de o ed o comple ing he sessions dec eases wi h
age, while he gende gap in o he ou comes does no signi ican ly change ac oss age
g oups. Las ly, we do no ind signi ican di e ences in he gende gaps in ou comes
by income le els o he municipali y o esidence.
This s udy con ibu es o he li e a u e on gende di e ences in educa ion in se e al
ways. Fi s , while p e ious esea ch has documen ed gende gaps in adi ional ace- o-
ace educa ion, he gende gap in online lea ning ou comes has been unde s udied. Ou
pape con ibu es o he li e a u e by p o iding e idence om a con empo a y online
lea ning ool.1This allows us o documen gende di e ences in comple ing he daily
exe cises, a he han he es ou comes, which is impo an in se e al ways. Bala
e al. (2018) show ha pe o mance in cogni i e es s, such as PISA, is in luenced by
non-cogni i e skills (also named as pe sonali y ai s, so skills, o cha ac e skills),
by using a decomposi ion in PISA es sco es. Also, Anaya e al. (2022) show ha he
di icul y le el a he beginning o es s may in luence he success o he la e ques ions.
The e is also some e idence in he psychology li e a u e sugges ing ha unde low-
s akes es ing condi ions, some indi iduals y ha de han o he s (Duckwo h e al.
2011). This issue becomes mo e impo an i he non-cogni i e e ec s di e by gende .
An example o his is he s udy by Mon olio and Tabe ne (2021), whe e hey ind ha
male uni e si y s uden s ou pe o m hei emale coun e pa s unde high p essu e.
Conside ing hese, i is essen ial o ocus on ou comes ha accumula e wi hin a mon h
as a esul o daily exe cises ha a e comple ed unde no o li le p essu e. In addi ion,
using da a om an online lea ning pla o m helps us documen gende gaps in s uden
e o and mo i a ion. A ending he es session, ime spen on each es i em, and
sel - epo ed e o ha e been used as p ima y measu es o mo i a ion in he p e ious
li e a u e documen ing gende gaps in he mo i a ion o s uden s, whe e emales a e
gene ally ound o exe mo e e o compa ed o males (DeMa s e al. 2013).2We add
o his li e a u e by analyzing he gende gap in e o when using an online lea ning
pla o m, whe e we measu e e o by he indica o s c ea ed based on he numbe o
comple ed online sessions in a mon h.
Second, unde s anding he sou ce o gende disc imina ion o bias wi hin he amily
is c ucial o designing ele an and e ec i e policies. Se e al s udies show ha mo h-
1Online lea ning has become he in e es o many esea che s since he s a o he Co id-19 ou b eak,
subsequen school closu es, and he ise o online educa ion. Among o he s, Che y e al. (2020) ind ha
child en expe ienced a educ ion in lea ning on a ma h lea ning pla o m used in US schools and Ikeda and
Yamaguchi (2021) ind ha school closu es du ing Co id-19 inc eased s uden s’ s udy ime using an online
lea ning se ice in Japan. Howe e , Che y e al. (2020) use school-le el da a, and hey do no examine
esul s by gende . On he o he hand, Ikeda and Yamaguchi (2021) do no ind any he e ogenous e ec by
gende .
2See DeMa s e al. (2013) o a comp ehensi e li e a u e e iew.
123
352 SERIEs (2024) 15:349–388
e s de o e mo e ime o childca e esponsibili ies han a he s, and his gap inc eased
a e he Co id-19 pandemic (Golin 2021; And ew e al. 2020; Del Boca e al. 2020).
We con ibu e o his pa o he li e a u e by compa ing he gende gap in educa-
ional ou comes ac oss wo ypes o amilies: he ones whe e online lea ning is mainly
supe ised by he mo he and hose by he a he . Al hough we canno iden i y he
causal impac o he di ision o amily esponsibili ies and childca e in ou se ing, we
desc ip i ely show whe he he gende gaps di e depending on he gende o he main
supe iso o he child en. Fu he mo e, ou s udy di e s om he p e ious li e a u e
by iden i ying he gende gap among siblings, while mos o he exis ing s udies ocus
on he gap among pee s. This enables us o es ima e he gende di e ences in lea ning
ou comes be ween a boy and a gi l aised in he same amily.
Thi d, we con ibu e o he ex ensi e li e a u e on he ela ionship be ween he
gende gap in educa ion and gende no ms. Se e al s udies ind ha coun ies wi h
mo e equal gende social no ms end o ha e smalle gende gaps in achie emen o
educa ional p e e ences (e.g., De San Román and De La Rica 2016; Nollenbe ge
e al. 2016; Rod íguez-Planas and Nollenbe ge 2018;Ge eke al.2020). Al hough
ou da a a e speci ic o one coun y, we adap his app oach in ou con ex o examine
whe he he gende gap in academic e o and pe o mance is la ge o smalle in
municipali ies wi h mo e egali a ian gende no ms, which we p oxy using he ela i e
sha es o emales’ and males’ employmen a es and con ibu ions o household cho es.
2 Da a and me hodology
2.1 Sma ick
We use indi idual-le el anonymous da a om Sma ick, an online lea ning pla o m
ha is used in o e 100 coun ies, including bu no limi ed o Spain, he UK, he
USA, Mexico, Colombia, Pe u, B azil, and Sou h A ica. The con en is o e ed in
di e en dialec s o Spanish, English, and Po uguese, acco ding o he use ’s choice.
In ou s udy, we use in o ma ion on membe s esiding in Spain since hey cons i u e
he la ges p opo ion o Sma ick use s.
Sma ick is one o he mos widely used online lea ning pla o ms in Spain. I is a
ma h lea ning ool ha o e s a 15-minu e online ma h session e e y day.3S uden s
engage in ou main a eas: men al calcula ion, easoning, logic, and p og amming. The
pla o m is designed o child en aged 4 o 14 and inco po a es a i icial in elligence
o c ea e pe sonalized sessions ailo ed o he s uden ’s knowledge and abili ies. These
sessions a e in e ac i e and guided, wi h each exe cise co ec ed immedia ely. I an
answe is inco ec , i explains how i should ha e been done co ec ly. The e o e,
3Al hough Sma ick also o e s eading exe cises, we ocus solely on he ou comes o ma h exe cises due
o da a es ic ions. Sma ick eleased he eading sec ion in Sep embe 2020. The e o e, we ha e a much
smalle sample o child en who comple ed he eading exe cises compa ed o ha o ma h, and we a e no
con iden ha he sample is la ge enough o de i e any conclusions. In addi ion, one o ou main ou comes
o measu e he speed o child en in comple ing he exe cises, namely he delayed comple ion a iable o
he eading sec ion was no p o ided o us in he o iginal da a om Sma ick.
123
SERIEs (2024) 15:349–388 353
pa en s’ assis ance is no equi ed and Sma ick encou ages child en o wo k on he
sessions independen ly.
Since Sma ick is mainly pa en -based, ei he he a he o he mo he egis e s he
child o he pla o m and he/she keeps ack o he child’s p og ess by checking he
daily emails sen by Sma ick ega ding he child’s a endance and pe o mance in he
sessions. In addi ion o e iewing he daily emails, pa en s can (and a e encou aged by
Sma ick o) log in o he pla o m using hei accoun o access a de ailed pe o mance
analysis and he child’s p og ess. This indi idual ela ionship es ablished be ween one
o he pa en s and Sma ick au oma ically assigns esponsibili y o he child’s online
lea ning p ocess o ha pa en .4
A e a 7-day ee ial pe iod, pa en s a e o e ed and choose om h ee ypes o paid
con ac s: one-mon h, h ee-mon h, and one-yea con ac s. We obse e egis a ion
da es o ee ial and paid con ac , as well as mon hly a e ages o he ou comes
o each child s a ing om hei con ac da e. We ocus only on he du a ion o he
i s con ac .5Ou da ase includes membe s egis e ed o a paid con ac be ween
Janua y 2019 and July 2021. We es ic ou sample o he membe s who egis e ed o
a ee ial pe iod s a ing om Janua y 2019 o obse e hei i s con ac ou comes,
as well as hose who egis e ed o a paid con ac be o e 20 June 2021, o ensu e ha
we obse e he ou comes o a leas one comple e mon h.
Ou da ase includes child en’s age, gende , and he p esence o a heal h condi-
ion.6We also ha e in o ma ion abou he municipali y o esidence.7Addi ionally,
we obse e he anonymous pa en ID, which allows us o iden i y he siblings in he
sample, as well as he gende o he pa en who egis e ed he child on he pla o m.
Since Sma ick is a paid pla o m, he cha ac e is ics o i s use s a e likely o di -
e om hose o he o e all Spanish popula ion, aising conce ns abou he ex e nal
alidi y o ou esul s. The e o e, ollowing Che y e al. (2020), we p esen he demo-
g aphic cha ac e is ics o Sma ick use s in ou sample in Table 1 o show he ex en
o he selec ion. Since we do no ha e in o ma ion on he cha ac e is ics o pa en s,
we p oxy he demog aphic cha ac e is ics using he income le els o he municipali-
ies whe e hey eside. Speci ically, we compa e he income qua iles o he Sma ick
municipali ies ha we de ine as hose whe e a leas 5 (and 1) Sma ick membe s li e,
o he income qua iles o all Spanish municipali ies. As we expec , he esul s e eal
a selec ed sample in e ms o income dis ibu ion.
4E en hough one pa en egis e s he child and ecei es he emails abou he lea ning p ocess, we only
use his ela ionship as a p oxy o “being he main supe iso ” since we canno obse e whe he he ac ual
supe iso is he one who egis e s on he pla o m.
5We do no include he ou comes o he mon hs o la e con ac s because we canno clea ly iden i y
whe he a missing alue co esponds o a mon h wi hou a paid con ac o a mon h wi h no comple ed
sessions.
6The heal h condi ions include high in ellec ual capaci y, dyscalculia, dyslexia, in ellec ual disabili y, hea -
ing disabili y, ce eb al palsy, ma u a ional delay, Down synd ome, a en ion de ici hype ac i i y diso de
(ADHD), au ism, Aspe ge ’s synd ome, and pe asi e de elopmen al diso de .
7The municipali y o esidence is de ec ed by Sma ick as he loca ion whe e hey i s egis e on he
pla o m. In some cases, we obse e di e en loca ions o child en egis e ed by he same pa en , o he
loca ion in o ma ion is missing o a child bu iden i ied o his/he sibling. In hese cases, we assign all
child en o a gi en pa en he i s iden i ied loca ion.
123
354 SERIEs (2024) 15:349–388
Mo eo e , he public school pa icipa ion a e o he Sma ick use s8is 49.4%, while
he public school pa icipa ion a e in he Spanish popula ion a he p ima y school
le el is 68%. O e all, Table 1shows he deg ee o selec ion in e ms o income le el
and public school pa icipa ion in ou sample compa ed o he Spanish popula ion. In
his selec ed sample, we an icipa e less gende bias since hese pa en s a e likely o be
mo e educa ed and mo e awa e o gende equali y conce ns compa ed o he a e age
Spanish popula ion. The e o e, we belie e ha ou esul s ep esen a lowe bound o
gende bias in his con ex .
The ou comes o in e es in ou s udy include he o al numbe o sessions comple ed
in a mon h, he a e age a io o ime spen o comple e he exe cises o he expec ed
ime, and he a e age a io o co ec ly sol ed p oblems. Using he o al numbe o
sessions comple ed in each mon h, we also c ea e wo bina y ou come a iables ep-
esen ing whe he he child comple es a leas one session and a leas wen y sessions
in each mon h, espec i ely. We ca ego ize hese ou come a iables in o wo g oups:
child en’s e o and ela i e pe o mance. The measu es o e o include pe sis ency,
comple ion, and sessions, while he measu es o ela i e pe o mance include delayed
comple ion and accu acy. Pe sis ency and comple ion a e bina y a iables ha ake
he alue 1 i a leas one session and wen y sessions a e comple ed in each mon h,
espec i ely. Sessions ep esen s he a e age numbe o sessions comple ed pe mon h.
Delayed comple ion is he a e age a io o ime spen on p oblems o he expec ed ime.
Las ly, accu acy is he a e age a io o co ec ly sol ed p oblems.
I is impo an o no e ha we only obse e he accu acy and he delayed com-
ple ion a iables o he mon hs wi h a leas one comple ed session. We cons uc
hese ou comes as a e ages o e he ully obse ed mon hs o he i s con ac .9Fo
consis ency, we exclude a mon hly ou come i we do no obse e he ull mon h in ou
da ase . Fo example, i he i s con ac is o 3 mon hs bu ou da a only co e he
pe iod o he i s mon h and a hal , we only conside he ou comes o he i s mon h.
The i s panel in Table 4shows he desc ip i e s a is ics in he ull sample, which
includes 28,236 child en esiding in Spain. O hese, 52% o hem a e gi ls, c ea -
ing a balanced sample in e ms o child gende . Howe e , he e is an imbalance in
pa en gende , wi h 66% o he pa en s being mo he s. This a io aligns wi h he di -
e ence in a e age minu es spen pe day wi h child en in eaching- ela ed ac i i ies
by mo he s and a he s. Acco ding o he na ionally ep esen a i e Spain Time Use
Su ey 2009–2010, mo he s and a he s spend an a e age o 5.26 and 2.68min pe
day, espec i ely, on eaching- ela ed ac i i ies wi h hei child en, indica ing ha 66%
o he o al eaching ime is con ibu ed by mo he s.10,11
8Acco ding o a su ey conduc ed by Sma ick in Sep embe 2021 on a ep esen a i e sample o Sma ick
membe s, consis ing o 2894 esponde s.
9We conduc ed a sepa a e analysis ocusing on he ou comes o he i s mon h, ega dless o he con ac
ype. The esul s a e e y simila o he baseline analysis and a e a ailable upon eques .
10 A mo he ( a he ) is de ined as a woman (man) who has a daugh e o a son li ing in he same household,
bu no a g andpa en .
11 Al hough we expec ou sample o be mo e egali a ian in e ms o gende no ms and he sha e o
esponsibili ies han he a e age popula ion because hey eside in municipali ies wi h highe han a e age
income, we obse e ha he a io o mo he s egis e ing hei child en o he pla o m is e y simila o
he mo he s’ sha e o ime de o ed o eaching ac i i ies wi h child en on a e age. On he o he hand,
123
SERIEs (2024) 15:349–388 355
Table 1 Compa ison o demog aphic cha ac e is ics be ween he Sma ick use s and he Spanish popula ion
Sma ick municipali ies Spanish popula ion
≥5membe s ≥1membe
A e age HH ne income
25 h pe cen ile 27,442 25,248 21,864
Median 31,142 28,862 25,535
75 h pe cen ile 34,978 33,519 30, 411
Public school pa icipa ion 49.40% 49.40% 68%
Numbe o people
Numbe o Sma ick membe s 26,648 28,236 –
2019 popula ion 31,952,187 38,733,271 47,026,208
Numbe o municipali ies 484 1302 8151
This able shows he demog aphic cha ac e is ics o Sma ick use s, ollowing Che y e al. (2020). Sma ick
municipali ies a e de ined as municipali ies whe e a leas 1 o 5 Sma ick membe s eside. Income and
popula ion in o ma ion is based on 2018 s a is ics om he Spanish S a is ical O ice (INE)
The egis e ed child en in ou da ase ange in age om 4 o 16, wi h an a e age
age o 8.54. App oxima ely 6% o he child en in ou sample ha e a heal h condi ion.
Since we do no ha e in o ma ion on household income le els, we use he a e age
household ne income in hei municipali y o esidence as a p oxy o income, wi h a
mean alue o 37,628 Eu os. A o al o 70% o ou sample egis e ed o he pla o m
a e he Co id-19 ou b eak in Spain (a e Ma ch 9, 2020), and 48% egis e ed in
he i s 3 mon hs o he ou b eak when he schools we e closed, and people had o
s ay a home. On a e age, each child has a o al o 4.28 con ac s, and 52% o he
i s con ac s ha e a du a ion o 1 mon h. An a e age Sma ick use comple es 21.47
sessions in a mon h. Fu he mo e, 95% o he use s comple e a leas one session,
and 56% comple e a leas 20 sessions each mon h du ing hei i s con ac pe iod.
Child en, on a e age, spend 1.12 imes he expec ed ime o comple e he sessions and
answe 84% o he exe cises co ec ly.
The second panel in Table 4p esen s desc ip i e s a is ics o he sibling subsample,
which consis s o 7533 child en who ha e a leas one sibling o he opposi e gende
egis e ed on he pla o m. We c ea e his subsample o analyze gende di e ences
among siblings, allowing us o con ol o amily o household ixed e ec s while
documen ing he gende gaps. The dis ibu ions o a iables in his sample closely
esemble hose in he ull sample. In he siblings’ subsample, he a e age numbe o
child en wi h he same pa en is inhe en ly highe , and we obse e ha 50% o he
obse a ions in his subsample a e om households whe e he eldes child is a boy.
Table OA1 in he online appendix shows he mean di e ences o he a iables
by child gende . In he ull sample, boys a e on a e age 0.23 yea s younge , 4%
less likely o be egis e ed by hei mo he s, and 3% mo e likely o ha e a heal h
we obse e ha in he Time Use Su ey 2009–2010, mo he s’ co esponding ime sha e in highe -income
amilies (mo e han 2000 Eu os ne household income pe mon h) is a ound 60%, and mo he s spend mo e
han hal o he o al ime spen in all income g oups. The e o e, we belie e ha he Sma ick use s a e no
e y di e en han he a e age popula ion in e ms o pa en al ime in es men o eaching- ela ed ac i i ies.
123
356 SERIEs (2024) 15:349–388
condi ion compa ed o gi ls. In he siblings’ subsample, simila ly, boys a e on a e age
0.25 yea s younge and 1% mo e likely o ha e a heal h condi ion han gi ls. Boys
in he siblings’ subsample comple es on a e age 0.36 mo e sessions han gi ls pe
mon h. In bo h samples boys and gi ls come om simila backg ounds in e ms o
income le el and gende -age composi ion. Fo he ela i e pe o mance a iables, we
obse e s a is ically signi ican gende gaps in a o o boys, wi hou con olling o
any cha ac e is ics; howe e , he magni udes a e small.
Table OA2 shows he mean di e ences by he gende o he pa en . In line wi h
he in o ma ion in he p e ious able, a he s a e 4% less likely o egis e gi ls on he
pla o m compa ed o he mo he s in he ull sample. On a e age, child en egis e ed
by hei a he s a e less likely o ha e a heal h condi ion and he numbe o child en
in hei amily is sligh ly lowe compa ed o hose egis e ed by hei mo he s, in bo h
he ull sample and siblings’ subsample. The p obabili y o he eldes child being a
boy is 2 pp lowe in he amilies whe e he a he egis e s child en in he siblings’
subsample. In he ull sample, child en egis e ed by hei a he s comple e 0.23 ewe
sessions pe mon h and hey ha e on a e age 2 pp lowe comple ion a e han hose
egis e ed by hei mo he s. Howe e , we do no obse e signi ican di e ences in he
ela i e pe o mance ou comes.
Figu e3p esen s he e olu ion o he o al numbe o new egis a ions pe mon h.
As indica ed by he ed line, he numbe o egis a ions inc eased disp opo iona ely
in he mon h when he Co id-19 ou b eak s a ed. Howe e , s a ing om June 2020,
he numbe s e u ned o p e ious le els. Figu es4and Fig.5show he o al numbe o
egis a ions pe mon h, ca ego ized by he gende o he child and by he gende o
he pa en , espec i ely. In e ms o pe cen ages, Fig.6does no show a clea pa e n
o gende pai s o e he mon hs. As indica ed in Fig.7, he majo i y o membe s a e
p ima y school s uden s, ep esen ing a cons an end. Howe e , he e is an inc ease
in egis a ions by p eschool child en along wi h a dec ease in egis a ions by sec-
onda y school s uden s o e ime. Figu e8shows he pe cen age o new egis a ions
by income ca ego ies, de ined based on he qua iles o a e age income in he munic-
ipali y o esidence. The majo i y o membe s eside in high-income municipali ies,
and his end emains s able o e ime.
These igu es sugges ha he e is no clea pa e n o change in he gende com-
posi ion o child en and pa en s, as well as in he age and income dis ibu ion o e
ime, including he pe iod a e he Co id-19 ou b eak. In o he wo ds, he obse able
cha ac e is ics o ou sample do no exhibi signi ican changes o e ime, excep o
a sligh shi in he age ca ego ies.12
12 We a e awa e ha he cha ac e is ics o new egis a ions in June 2021 seem di e en han in p e ious
mon hs. One po en ial eason is ha we only include new egis a ions un il June 21 since we es ic ou
sample o he ully obse ed mon hs (see Sec . 2.1 o a de ailed explana ion), and ela ed o his, we only
obse e 12 new egis a ions in his mon h.
123
SERIEs (2024) 15:349–388 363
3.3 Di e ences by age and income le els
We hen analyze whe he gende gaps in e o and ela i e pe o mance ou comes
di e by he age o he child en. To do so, we u ilize bo h he con inuous age a iable
and he ca ego ical age a iable ha we de ine acco ding o he school le els (p eschool
(age 4–5), p ima y school (6–11), and seconda y school (12–16)). We in e ac he age
a iable wi h he gi l dummy a iable in he ull sample and es ima e he eg ession
condi ional on he gende o he pa en who supe ises he child.
In Table 10, Panel A shows ha while he gende gap in he e o ou comes and
accu acy a iable does no change signi ican ly by age, he gende gap in delayed
comple ion dec eases wi h age. In Table 10, Panel B shows ha he gende gap in
comple ion ou comes p ima ily comes om p eschool kids and we obse e a simila
pa e n in e ms o he magni udes o sessions ou come, al hough he es ima ions a e
no p ecise. While he accu acy gap comes om he p ima y school kids, he delayed
comple ion gap is mo e p onounced o child en in bo h p eschool and p ima y school.
O e all, Table 10 sugges s ha he gende gaps a e s onge o younge gi ls, bu mos
o hem ade away wi h age.
Nex , we c ea e income ca ego ies wi hin he Spanish popula ion and among
Sma ick use s based on he qua iles o a e age income in he municipali y o es-
idence. We hen in e ac his ca ego ical a iable wi h he gi l dummy a iable o
explo e whe he he sign and magni ude o he gende gaps di e depending on he
income le el. As shown in Tables 11 and 12, we do no obse e any signi ican di -
e ences in he gende gaps by income ca ego ies. We belie e ha he main eason o
his is he high homogenei y o ou sample in e ms o income le el, as he majo i y
o he use s come om high-income amilies.22
3.4 Robus ness checks
One conce n in he con ex o a paid lea ning pla o m is he po en ial nega i e selec ion
on abili y in ou sample. In o he wo ds, amilies migh be mo e likely o egis e hei
child en on his pla o m i hei child en a e no pe o ming well academically, and his
selec ion migh also a y by he gende o he child, which could po en ially a ec
ou esul s. To add ess his conce n, we epea he analysis in he “co id sample”,
which we de ine as he sample o child en who egis e ed on he pla o m be ween
Ma ch 9 and May 31, 2020. This pe iod co esponds o he i s wa e o he Co id-19
ou b eak in Spain. Du ing his pe iod, pa en s we e seeking educa ional ools o help
hei child en s ay on ack because schools we e closed unexpec edly, and dis ance
22 We p esen he he e ogenei y analysis by he gende o he eldes sibling, age, income a he municipali y
le el and gende no ms measu es a he municipali y le el in he ull sample only. We do no un he
he e ogenei y analysis by gende no ms and income in he siblings’ subsample and including he pa en
ixed e ec s. The main eason is ha in he siblings’ subsample many municipali ies ha e a ew amilies wi h
a leas one gi l and one boy using he Sma ick pla o m, making i di icul o es ima e he he e ogenei y
by municipali y-le el in o ma ion, wi hin amilies. We also canno un he he e ogenei y analysis by gende
o he eldes sibling in he siblings’ subsample and including pa en ixed e ec s, since he e is no a ia ion
in he gende o he eldes sibling o a gi en pa en . Howe e , we explo e he he e ogenei y by age in he
siblings’ subsample by including pa en ixed e ec s. The esul s a e in line wi h hose in he ull sample
and a e a ailable upon eques .
123
364 SERIEs (2024) 15:349–388
educa ion did no s a immedia ely a e he school closu es. The e o e, we belie e
ha po en ial selec ion on abili y and gende in egis a ions plays a minimal ole in his
ime pe iod. This idea is suppo ed by he inc ease in he o al numbe o egis a ions
and di e ences in he cha ac e is ics and ou come a iables o he child en egis e ed
du ing his pe iod, as shown in Fig.3and Table OA3.
When we compa e he cha ac e is ics and ou come a iables o child en in he
co id sample o he es o he sample (as shown in Table OA3), we ind some no able
di e ences. In his sample, he sha e o gi ls egis e ed o he pla o m is 3% lowe ,
while he sha e o mo he s is 4% highe . The child en in he co id sample a e, on
a e age, abou hal a yea younge , less likely o ha e a heal h condi ion and li e in
municipali ies wi h, on a e age, lowe income le els. While in he ull sample, he
a e age numbe o child en in he amilies is sligh ly highe , in he siblings’ sample i
is sligh ly lowe . In addi ion o he cha ac e is ics, we also obse e di e ences in he
dis ibu ion o ou come a iables. Child en in he co id sample comple e, on a e age,
2.27 mo e sessions pe mon h, a e 1% mo e likely o be pe sis en and 12% mo e
likely o comple e a leas 20 sessions pe mon h compa ed o he o he s. They also
spend less ime on comple ing he sessions and sol e sligh ly mo e exe cises co ec ly.
These di e ences sugges ha i he e is any selec ion on abili y in ou sample, his
selec ion would be he smalles in his pe iod. Fu he mo e, he income le el o he
municipali ies whe e he child en li e is close o he a e age Spanish popula ion
compa ed o he es o he sample.
We es ima e he gende gaps in he “co id sample” o examine whe he ou esul s
a e a ec ed by a po en ial selec ion on abili y and gende . As shown in Fig.9and
Fig.10, he magni udes o es ima ed gende gaps in he co id sample a e e y simila
o hose in ou baseline analysis. Howe e , he es ima es a e no as p ecise as he
baseline es ima es in he siblings’ subsample, which migh be due o he lowe numbe
o obse a ions. The magni udes o he coe icien es ima es a e p esen ed in Tables
13 and 14.
Nex , we di ide he ull sample in o wo g oups: hose who egis e ed be o e and
a e he Co id-19 ou b eak in Spain. In Table 15, we p esen he eg ession esul s
whe e we in e ac he gi l dummy a iable wi h he a he dummy a iable in hese
wo subsamples.
We ind ha he gende gap in he pe sis ency ou come and i s nega i e associa ion
wi h he a he s’ supe ision come om he egis a ions made be o e he ou b eak.
Al hough he signs o he es ima es a e mos ly in he same di ec ion in he sample
o egis a ions a e he ou b eak o he e o ou comes, hey a e no signi ican ly
di e en om ze o. Rega ding he ela i e pe o mance ou comes, we obse e ha in
he sample o egis a ions a e he ou b eak, he gende gap in he delayed comple ion
ou come is sligh ly lowe o he child en wi h a he s.
We u he examine whe he he es ima ed gende gaps in he siblings’ subsample
a e obus o including p o ince ixed e ec s ins ead o pa en ixed e ec s in he
eg essions, as shown in Table 16. While he magni udes o he es ima es sligh ly
change, hey emain e y simila o he baseline es ima es and main ain he same
s a is ical signi ican le els.
Fo he bina y ou come a iables (pe sis ency and comple ion), we es ima e logi
eg essions as an al e na i e o he linea p obabili y model. Table 17 p esen s he
123
SERIEs (2024) 15:349–388 365
esul s o he ull sample and siblings’ sample, espec i ely. The es ima ed gende
gaps a e consis en in di ec ion wi h hose es ima ed using linea p obabili y models.
Nex , we exclude child en wi h a heal h condi ion om he sample o assess whe he
ou esul s a e a ec ed by he p esence o a heal h condi ion. Table 18 shows he
es ima ed gende gaps in he ull sample o child en wi hou a heal h condi ion. The
esul s closely esemble he baseline esul s epo ed in Table 7.
We cons uc pe sis ency and comple ion a iables based on he numbe o sessions
comple ed pe mon h. We ind ha gi ls a e 0.7% mo e likely o comple e a leas one
session pe mon h (pe sis ency) han boys when he main supe iso is he mo he .
Howe e , we do no obse e gende gaps in he comple ion a iable (see Table 6).
To assess he s abili y o ou esul s o di e en cu -o s o he numbe o sessions
comple ed pe mon h, we cons uc addi ional ou come a iables wi h a ious cu -o s
(1, 6, 11,..., 26), whe e he ou come o he i s cu -o is he same as he pe sis ency
a iable. The esul s a e p esen ed in Fig.11, whe e we es ima e Eq.1 o di e en
cu -o ou comes. In he mo he s’ sample, he e a e no signi ican gende gaps o
di e en cu -o s, excep o he cu -o o 1. Howe e , in he a he s’ sample, we
obse e ha gi ls a e 1.8% (2%) less likely han boys o comple e a leas 11 sessions
(16 sessions) pe mon h.
In ou main speci ica ion, we use he con inuous age a iable o con ol o he
di e ences in he gende gap ac oss di e en ages. Addi ionally, in Fig.12, we p esen
ou main esul s by including he age ixed e ec s ins ead o he con inuous age
a iable. The signi icance le els and he magni udes o he coe icien es ima es in
his model a e e y simila o hose o ou main speci ica ion.
To u he con ol o di e ences by age, we epea ou analysis by no malizing all
he ou come a iables wi hin each age. This is pa icula ly impo an o he siblings’
sample, whe e we compa e siblings a di e en ages, and he di icul y le el o exe -
cises o di e en ages is likely o be di e en . The esul s p esen ed in Figs. 13 and
14 a e e y simila o ou baseline esul s. I any hing, we obse e la ge magni udes
o coe icien es ima es o he pe sis ency a iable, wi h he signs emaining he same
as in he baseline.
3.5 Po en ial mechanisms and discussion
Ou esul s sugges gende gaps in online ma h lea ning in a o o boys in pe o -
mance ou comes. Fo he e o ou comes, we do no ind gende gaps in he ull sample
on a e age; howe e , we ind a gende gap in a o o boys o he sessions a iable
when we ocus on he siblings’ subsample. Fu he analyzing he gende di e ences
in he mo he s’ and a he s’ samples, we ind ha in he ull sample, he gende gap in
he pe sis ency a iable is posi i e o hose supe ised by hei mo he s bu nega i e
o hose supe ised by hei a he s, wi h a signi ican di e ence obus o Benjamini
and Hochbe g (1995) mul iple hypo hesis es ing. Howe e , his di e ence is no s a-
is ically signi ican in he siblings’ subsample and is no obse ed o o he a iables
measu ing e o .23 Fo his eason, we in e p e ou indings as sugges i e e idence
23 Al hough we ind a signi ican di e ence in he gende gap by he gende o he pa en in he ull sample
o he sessions a iable, his di e ence is no obus o mul iple hypo hesis es ing.
123
366 SERIEs (2024) 15:349–388
o na owe /posi i e gende gaps in e o ou comes when he main supe iso is he
mo he . While da a limi a ions p e en us om analyzing he mechanisms behind hese
esul s, in his sec ion, we discuss he po en ial explana ions o ou indings.
In addi ion o he be e ma h pe o mance o boys, which is also e iden in adi-
ional educa ion, ano he con ibu ing ac o may be ha boys end o be mo e in e es ed
in ideo games compa ed o gi ls. This in e es could explain di e ences in e o ou -
comes, as boys migh be mo e mo i a ed o comple e sessions hey ind enjoyable.
Howe e , i is impo an o no e ha we do no obse e his pa e n o child en wi h
hei mo he s as he main supe iso s. This di e ence becomes mo e p onounced when
examining he p obabili y o comple ing a leas 1, 11, and 16 sessions pe mon h.
One possible explana ion o ou sugges i e esul s on he di e ences in e o
gende gaps by he gende o he main supe iso is ha mo he s migh p o ide
g ea e suppo o hei daugh e s in engaging wi h ma h lea ning ac i i ies compa ed
o a he s, which could lead o he disappea ance o he di e ence in e o gaps. This
may also explain why we do no obse e such di e ences in pe o mance gaps.
Since Sma ick is a pla o m whe e pa en al assis ance is no equi ed du ing he
sessions, pa en s migh mainly in luence he child’s decision o s a he exe cises.
Howe e , due o da a limi a ions, we canno di ec ly obse e he ac ual suppo p o-
ided by pa en s. Bo h pa en s may p o ide suppo e en i only one has egis e ed he
child. The e o e, we canno ule ou he possibili y ha he di e ences in gende gaps
depending on he gende o he supe iso migh be d i en by di e ences in amily
and household cha ac e is ics ha a e no obse able o us.
One migh expec ha in mo e gende -egali a ian municipali ies, a he s a e mo e
likely o supe ise child en. This expec a ion could c ea e ension in ou in e p e a ion
o he esul s since we ind (I) sugges i e e idence o la ge gende gaps in e o when
he a he is he main supe iso , and (ii) li ing in a municipali y wi h mo e egali a ian
gende no ms is associa ed wi h posi i e gende gaps in e o ou comes in a o
o gi ls. To in es iga e his, we es ima e a linea eg ession o he dummy a iable,
which akes he alue o 1 i he mo he supe ises, on he gende no m measu es we
cons uc (as de ailed in Sec . 3.2). We con ol o he gende o he child and include
o he con ol a iables in ou main speci ica ion (Eq.1). Table 19 shows ha he e is
no e idence o suppo a highe likelihood o a he s’ supe ision in mo e egali a ian
municipali ies.
Finally, i is impo an o emphasize ha ou esul s a e speci ic o ma hema ics
aining. Gi en ha pe o mance and e o by gende ha e been obse ed o di e
signi ican ly ac oss adi ionally male o emale subjec s, he ou comes may a y when
examining online lea ning in o he subjec s.
4 Conclusion
This pape in es iga es gende gaps in e o and ela i e pe o mance ou comes in
he con ex o online educa ion, ocusing on child en p ima ily supe ised by hei
mo he s and a he s. U ilizing da a om a widely used online lea ning pla o m in
Spain, we ind signi ican gende gaps in pe o mance ou comes in a o o boys, and
hese esul s hold when we compa e siblings o he opposi e gende . Mo eo e , we ind
123
SERIEs (2024) 15:349–388 367
sugges i e e idence ha he gende gaps in he e o ou comes a e mo e p onounced
o he child en who a e supe ised by hei a he s.
Since ou s udy ocuses on he con ex o online lea ning, ou ou come a iables
may no di ec ly co espond o hose commonly used in p e ious li e a u e, such
as s anda dized es sco es. Among ou ou come a iables, he mos compa able o
s anda dized es sco es is accu acy, which ep esen s he a e age a io o co ec ly
sol ed p oblems when using he online pla o m o each child. We obse e an accu acy
gende gap o 8% o a s anda d de ia ion24 a o ing boys. This es ima ed gap is o
a simila magni ude o he ma h gende gap in PISA 2018, whe e boys ou pe o med
gi ls wi h a di e ence o 6.8% o a s anda d de ia ion.25
We ind ha li ing in a municipali y wi h mo e egali a ian gende no ms is asso-
cia ed wi h gende gaps in e o ou comes in a o o gi ls, which is in line wi h he
p e ious indings in he li e a u e (e.g., by De San Román and De La Rica 2016;Nol-
lenbe ge e al. 2016; Rod íguez-Planas and Nollenbe ge 2018;Ge eke al.2020).
Howe e , he di e ence in e o ou comes does no ansla e in o he gende gap in
ela i e pe o mance ou comes.
We con ibu e o he exis ing li e a u e on gende gaps in lea ning ou comes by
p o iding new e idence om an online lea ning con ex while mos p e ious s udies
ocus on adi ional educa ion se ings.
The inc easing adop ion o online lea ning echnologies by bo h pa en s and schools
sugges s ha hese ools will become an impo an pa o egula educa ion sys ems
in he nea u u e. In ligh o hese de elopmen s, we belie e ha i is e y impo an
o documen gende di e ences in he use o hese ools, as hese a e likely o ansla e
in o di e ences in u he educa ion and labo ma ke ou comes.
While ou analysis ocuses on Spain, i is impo an o no e ha he Sma ick
pla o m is u ilized by child en in many o he coun ies, wi h he op ion o h ee
di e en languages. The e o e, we belie e ha ou s udy is ele an no only in he
Spanish con ex bu also in he con ex o o he coun ies.
Supplemen a y In o ma ion The online e sion con ains supplemen a y ma e ial a ailable a h ps://doi.
o g/10.1007/s13209-024-00299-1.
Decla a ions
Con lic o in e es None.
Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which
pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e
app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons licence,
and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included
in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he ma e ial. I
ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed
by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he
copy igh holde . To iew a copy o his licence, isi h p://c ea i ecommons.o g/licenses/by/4.0/.
24 In he siblings’ subsample, whe e we compa e opposi e-gende siblings, he co esponding gende gap
is 5.4% o a s anda d de ia ion.
25 Acco ding o he au ho s’ own calcula ions based on PISA 2018 desc ip i e s a is ics epo ed in OECD
(2019).
123
368 SERIEs (2024) 15:349–388
Appendix
See Tables 4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,Figs.3,4,5,6,7,8,
9,10,11,12,13 and 14.
Table 4 Desc ip i e s a is ics
Va iable N Mean S d. De . Min Max
1. Full sample
Female 28236 0.52 0.5 0 1
Mo he 28236 0.66 0.48 0 1
Age 28236 8.54 2.33 4 16
Heal h condi ion 28236 0.06 0.24 0 1
Income 27985 37628 9587 16692 90902
Regis e ed a e co id 28236 0.7 0.46 0 1
Co id sample 28236 0.48 0.5 0 1
Numbe o child en 28236 1.72 0.88 1 9
To al con ac s 28236 4.28 4.16 1 31
Fi s con ac 28236
1 14615 52%
3 10806 38%
12 2815 10%
Sessions 28236 21.47 7.91 0 32
Pe sis ency 28236 0.95 0.22 0 1
Comple ion 28236 0.56 0.5 0 1
Delayed comple ion 28105 1.12 0.28 0.3 3.26
Accu acy 28116 0.84 0.06 0.22 1
2. Siblings’ sample
Female 7533 0.5 0.5 0 1
Mo he 7533 0.65 0.48 0 1
Age 7533 8.56 2.37 4 16
Heal h condi ion 7533 0.05 0.22 0 1
Income 7459 38137 9987 17354 90902
Regis e ed a e co id 7533 0.71 0.45 0 1
Co id sample 7533 0.51 0.5 0 1
Numbe o child en 7533 2.45 0.83 2 9
Eldes is a boy 7533 0.5 0.5 0 1
To al con ac s 7533 4.36 4.29 1 30
Fi s con ac 7533
1 3926 52%
3 2782 37%
12 825 11%
Sessions 7533 22.14 7.76 0 32
Pe sis ency 7533 0.95 0.21 0 1
123
SERIEs (2024) 15:349–388 369
Table 4 con inued
Va iable N Mean S d. De . Min Max
Comple ion 7533 0.58 0.49 0 1
Delayed comple ion 7502 1.09 0.26 0.3 2.84
Accu acy 7505 0.84 0.06 0.42 1
The able shows he desc ip i e s a is ics o he ull sample and siblings’ sample. The income a iable
is measu ed as he a e age household ne income o he municipali y o esidence, and he co id sample
co esponds o he sample o use s who egis e ed be ween Ma ch 9, 2020, and May 30, 2020
Table 5 Gende gaps in he ull sample and siblings’ sample
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Full sample
Gi l 0.003 −0.002 0.001 −0.080∗∗∗ 0.148∗∗∗
(0.003)(0.005)(0.014)(0.009)(0.011)
[0.25][0.77][0.97][0.00][0.00]
[0.41] [0.97] [0.97] [0.00] [0.00]
Mo he −0.005∗0.011∗∗ 0.024∗∗ −0.048∗∗∗ 0.007
(0.003)(0.005)(0.012)(0.014)(0.010)
Age −0.004∗∗∗ 0.002∗0.004∗−0.124∗∗∗ −0.068∗∗∗
(0.000)(0.001)(0.003)(0.004)(0.002)
Heal h Cond. −0.015∗−0.017∗−0.044∗∗ −0.174∗∗∗ 0.346∗∗∗
(0.007)(0.008)(0.021)(0.033)(0.026)
Num o con 0.004∗∗∗ 0.023∗∗∗ 0.058∗∗∗ 0.008∗∗∗ 0.002
(0.000)(0.001)(0.002)(0.001)(0.002)
1s C:12 −0.143∗∗∗ −0.241∗∗∗ 0.092∗∗∗ −0.357∗∗∗ 0.538∗∗∗
(0.007)(0.011)(0.022)(0.019)(0.017)
1s C:3 −0.056∗∗∗ −0.205∗∗∗ −0.044∗∗ −0.118∗∗∗ 0.112∗∗∗
(0.004)(0.013)(0.020)(0.012)(0.013)
Cons an 0.985∗∗∗ 0.476∗∗∗ −0.418∗∗∗ 1.472∗∗∗ 0.631∗∗∗
(0.013)(0.039)(0.081)(0.103)(0.051)
P o ince FE Yes Yes Yes Yes Yes
Con . Mon h FE Yes Yes Yes Yes Yes
R20.068 0.144 0.107 0.121 0.092
Num. obs. 28236 28236 28236 28116 28105
N Clus e s 50 50 50 50 50
Siblings’ sample
Gi l 0.002 −0.013 −0.041∗∗∗ −0.054∗∗∗ 0.181∗∗∗
(0.003)(0.008)(0.011)(0.016)(0.020)
[0.53][0.11][0.00][0.00][0.00]
[0.53] [0.13] [0.00] [0.00] [0.00]
Age −0.005∗∗∗ −0.007∗∗∗ −0.025∗∗∗ −0.037∗∗∗ −0.031∗∗∗
(0.001)(0.002)(0.004)(0.011)(0.011)
123
370 SERIEs (2024) 15:349–388
Table 5 con inued
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Heal h Cond. −0.016 −0.022 −0.028 −0.348∗∗∗ 0.276∗∗
(0.012)(0.029)(0.064)(0.083)(0.127)
Cons an 1.069∗∗∗ 1.096∗∗∗ 0.975∗∗∗ 0.162 −0.650∗∗∗
(0.018)(0.032)(0.050)(0.141)(0.142)
Pa en FE Yes Yes Yes Yes Yes
Con . Mon h FE No No No No No
R20.718 0.814 0.869 0.685 0.635
Num. obs. 7533 7533 7533 7505 7502
N Clus e s 50 50 50 50 50
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e (in pa en hesis) clus e ed a he p o ince
le el. O iginal p- alues and Benjamini and Hochbe g adjus ed p- alues (i alic) a e gi en in b acke s. The
coe icien es ima es a e om he eg essions o i e di e en ou come a iables gi en in di e en columns,
on he indica o a iables o he child’s gende , pa en ’s gende , and he con ol a iables shown in he able
Table 6 Gende gaps in mo he s’ and a he s’ ( ull) sample
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Mo he s’ sample
Gi l 0.007∗0.004 0.018 −0.078∗∗∗ 0.157∗∗∗
(0.003)(0.007)(0.021)(0.014)(0.010)
[0.05][0.55][0.41][0.00][0.00]
[0.09] [0.55] [0.51] [0.00] [0.00]
R20.069 0.151 0.115 0.123 0.098
Num. obs. 18516 18516 18516 18435 18430
N Clus e s 50 50 50 50 50
Fa he s’ sample
Gi l −0.004 −0.013 −0.032∗−0.088∗∗∗ 0.131∗∗∗
(0.003)(0.008)(0.016)(0.015)(0.020)
[0.12][0.11][0.06][0.00][0.00]
[0.12] [0.12] [0.09] [0.00] [0.00]
R20.074 0.141 0.103 0.126 0.088
Num. obs. 9720 9720 9720 9681 9675
N Clus e s 50 50 50 50 50
Two sample - es
Di e ence 0.011 0.017 0.05 0.01 0.03
p- alue (0.01) (0.11) (0.07) (0.63) (0.24)
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e (in pa en hesis) clus e ed a he p o ince
le el. O iginal p- alues and Benjamini and Hochbe g adjus ed p- alues (i alic) a e gi en in b acke s. Each
eg ession includes a cons an e m, p o ince, and con ac yea -mon h ixed e ec s. Con ol a iables: age,
heal h condi ion, numbe o con ac s, ype o he i s con ac
123
SERIEs (2024) 15:349–388 371
Table 7 Gende gaps in he ull sample
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Gi l 0.007∗0.005 0.017 −0.078∗∗∗ 0.157∗∗∗
(0.003)(0.007)(0.021)(0.014)(0.011)
[0.05][0.54][0.41][0.00][0.00]
[0.09] [0.54] [0.51] [0.00] [0.00]
Fa he 0.011∗∗∗ −0.002 0.000 0.051∗∗ 0.005
(0.004)(0.007)(0.020)(0.022)(0.014)
[0.01][0.71][1.00][0.03][0.71]
[0.04] [0.89] [1.00] [0.06] [0.89]
Gi lxFa he −0.011∗∗∗ −0.018 −0.049∗−0.007 −0.025
(0.004)(0.011)(0.027)(0.023)(0.017)
[0.01][0.12][0.08][0.76][0.14]
[0.04] [0.17] [0.17] [0.76] [0.17]
Age −0.004∗∗∗ 0.002∗0.004∗−0.124∗∗∗ −0.068∗∗∗
(0.000)(0.001)(0.003)(0.004)(0.002)
Heal h Cond. −0.014∗−0.016∗−0.044∗∗ −0.173∗∗∗ 0.346∗∗∗
(0.007)(0.008)(0.022)(0.033)(0.026)
Num o con 0.004∗∗∗ 0.023∗∗∗ 0.058∗∗∗ 0.008∗∗∗ 0.002
(0.000)(0.001)(0.002)(0.001)(0.002)
1s C:12 −0.143∗∗∗ −0.241∗∗∗ 0.092∗∗∗ −0.357∗∗∗ 0.538∗∗∗
(0.007)(0.012)(0.022)(0.019)(0.017)
1s C:3 −0.056∗∗∗ −0.205∗∗∗ −0.044∗∗ −0.118∗∗∗ 0.112∗∗∗
(0.004)(0.013)(0.020)(0.012)(0.013)
Cons an 0.978∗∗∗ 0.484∗∗∗ −0.402∗∗∗ 1.423∗∗∗ 0.633∗∗∗
(0.014)(0.039)(0.087)(0.092)(0.047)
P o ince FE Yes Yes Yes Yes Yes
Con . Mon h FE Yes Yes Yes Yes Yes
R20.068 0.144 0.107 0.121 0.092
Num. obs. 28236 28236 28236 28116 28105
N Clus e s 50 50 50 50 50
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e (in pa en hesis) clus e ed a he p o ince
le el. O iginal p- alues and Benjamini and Hochbe g adjus ed p- alues (i alic) a e gi en in b acke s. The
coe icien es ima es a e om he eg essions o i e di e en ou come a iables gi en in di e en columns,
on he indica o a iables o he child’s gende , pa en ’s gende , hei in e ac ion, and he con ol a iables
123
372 SERIEs (2024) 15:349–388
Table 8 Gende gaps in mo he s’ and a he s’ (siblings’) sample
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Mo he s’ sample
Gi l 0.006 −0.002 −0.023∗∗ −0.063∗∗∗ 0.189∗∗∗
(0.005)(0.007)(0.011)(0.022)(0.026)
[0.22][0.75][0.04][0.01][0.00]
[0.27] [0.75] [0.06] [0.02] [0.00]
R20.703 0.824 0.870 0.687 0.634
Num. obs. 4889 4889 4889 4868 4867
N Clus e s 49 49 49 49 49
Fa he s’ sample
Gi l −0.006 −0.032∗−0.074∗∗ −0.038 0.165∗∗∗
(0.006)(0.019)(0.032)(0.025)(0.041)
[0.27][0.09][0.02][0.13][0.00]
[0.27] [0.16] [0.06] [0.17] [0.00]
R20.753 0.796 0.868 0.684 0.636
Num. obs. 2644 2644 2644 2637 2635
N Clus e s 50 50 50 50 50
Two sample - es
Di e ence 0.012 0.030 0.051 -0.025 0.023
p- alue (0.10) (0.13) (0.13) (0.45) (0.63)
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e (in pa en hesis) clus e ed a he p o ince
le el. O iginal p- alues and Benjamini and Hochbe g adjus ed p- alues (i alic) a e gi en in b acke s. Each
eg ession includes a cons an e m, p o ince, and con ac yea -mon h ixed e ec s. Con ol a iables: age,
heal h condi ion, numbe o con ac s, ype o he i s con ac
Table 9 Gende gap—di e ences by con inuous indica o s o gende no ms
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Employmen gende a io
Gi l 0.003 −0.011 −0.019 −0.078∗∗∗ 0.143∗∗∗
(0.006)(0.008)(0.019)(0.015)(0.019)
[0.62][0.16][0.34][0.00][0.00]
[0.62] [0.27] [0.42] [0.00] [0.00]
Em0.002 −0.011 −0.051∗0.005 −0.013
(0.005)(0.016)(0.028)(0.012)(0.017)
[0.74][0.50][0.08][0.68][0.46]
[0.74] [0.74] [0.38] [0.74] [0.74]
Gi lxEm0.002 0.016∗0.033 0.003 0.010
(0.007)(0.009)(0.027)(0.018)(0.019)
[0.80][0.07][0.22][0.87][0.60]
[0.87] [0.34] [0.56] [0.87] [0.87]
123
SERIEs (2024) 15:349–388 379
Table 16 Gende gaps in he mo he s’ and a he s’ (siblings’) sample—p o ince FE
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Mo he s’ sample
Gi l 0.005 −0.003 −0.028∗∗∗ −0.047∗∗∗ 0.205∗∗∗
(0.004)(0.005)(0.009)(0.016)(0.020)
[0.24][0.54][0.00][0.01][0.00]
[0.31] [0.54] [0.01] [0.01] [0.00]
R20.097 0.204 0.143 0.104 0.121
Num. obs. 4889 4889 4889 4868 4867
N Clus e s 49 49 49 49 49
Fa he s’ sample
Gi l −0.005 −0.027∗−0.066∗∗ −0.015 0.189∗∗∗
(0.004)(0.015)(0.027)(0.019)(0.029)
[0.21][0.07][0.02][0.43][0.00]
[0.26] [0.12] [0.05] [0.43] [0.00]
R20.108 0.196 0.149 0.128 0.126
Num. obs. 2644 2644 2644 2637 2635
N Clus e s 50 50 50 50 50
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e (in pa en hesis) clus e ed a he p o ince
le el. O iginal p- alues and Benjamini and Hochbe g adjus ed p- alues (i alic) a e gi en in b acke s. Each
eg ession includes a cons an e m, p o ince, and con ac yea -mon h ixed e ec s. Con ol a iables: age,
heal h condi ion, numbe o con ac s, ype o he i s con ac
Table 17 Logi eg essions
Mo he s Fa he s
Pe sis ency Comple ion Pe sis ency Comple ion
A. Full sample
Gi l 0.136 0.022 −0.117∗−0.062∗
(0.086)(0.034)(0.069)(0.037)
AIC 5723.798 22488.430 2975.372 12078.474
BIC 6389.041 23153.673 3585.837 12688.939
Log Likelihood −2776.899 −11159.215 −1402.686 −5954.237
De iance 5553.798 22318.430 2805.372 11908.474
Num. obs. 18516 18516 9720 9720
B. Siblings’ sample
Gi l 0.423 −0.079 −0.852 −0.681∗∗
(0.277)(0.124)(0.574)(0.320)
123
380 SERIEs (2024) 15:349–388
Table 17 con inued
Mo he s Fa he s
Pe sis ency Comple ion Pe sis ency Comple ion
AIC 4649.769 5455.443 2476.539 3073.614
BIC 18542.024 19347.699 9362.076 9959.150
Log Likelihood −185.884 −588.722 −67.270 −365.807
De iance 371.769 1177.443 134.539 731.614
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e (in pa en hesis) clus e ed a he p o ince le el.
Con ol a iables o he eg essions in Panel A: age, heal h condi ion, numbe o con ac s, ype o he i s
con ac , p o ince, and con ac yea -mon h ixed e ec s. Con ol a iables o he eg essions in Panel B:
age, heal h condi ion, and pa en ixed e ec s
Table 18 Gende gaps in he ull sample o child en wi hou a heal h condi ion
Pe sis ency Comple ion Sessions Accu acy Delayed c.
Gi l 0.008∗0.004 0.019 −0.071∗∗∗ 0.159∗∗∗
(0.004)(0.008)(0.025)(0.015)(0.012)
[0.06][0.60][0.46][0.00][0.00]
[0.10] [0.60] [0.57] [0.00] [0.00]
Fa he 0.011∗∗∗ −0.004 0.001 0.041∗0.006
(0.004)(0.008)(0.024)(0.023)(0.015)
[0.01][0.64][0.98][0.08][0.70]
[0.04] [0.88] [0.98] [0.19] [0.88]
Gi lxFa he −0.011∗∗ −0.016 −0.046 0.009 −0.025
(0.005)(0.013)(0.033)(0.025)(0.019)
[0.04][0.20][0.17][0.73][0.19]
[0.22] [0.26] [0.26] [0.73] [0.26]
R20.067 0.144 0.107 0.117 0.091
Num. obs. 26476 26476 26476 26373 26363
N Clus e s 50 50 50 50 50
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e (in pa en hesis) clus e ed a he p o ince
le el. O iginal p- alues and Benjamini and Hochbe g adjus ed p- alues (i alic) a e gi en in b acke s. Each
eg ession includes a cons an e m, p o ince, and con ac yea -mon h ixed e ec s. Con ol a iables: age,
numbe o con ac s, ype o he i s con ac
123
SERIEs (2024) 15:349–388 381
Table 19 Full sample— egis a ion p obabili y by gende no ms municipali ies
Mo he Mo he
Cho es −0.005
(0.007)
EmpGendRa io 0.005
(0.013)
Gi l 0.044∗∗∗ 0.044∗∗∗
(0.011)(0.011)
Age −0.002 −0.002
(0.001)(0.001)
Heal h cond. 0.073∗∗∗ 0.074∗∗∗
(0.011)(0.011)
Num o con −0.004∗∗∗ −0.004∗∗∗
(0.001)(0.001)
1s C:12 −0.086∗∗∗ −0.086∗∗∗
(0.011)(0.011)
1s C:3 −0.010 −0.010
(0.009)(0.009)
Cons an 1.519∗∗∗ 1.514∗∗∗
(0.032)(0.032)
P o ince FE Yes Yes
Con . Mon h FE Yes Yes
R20.013 0.013
Num. obs. 24896 24896
N Clus e s 50 50
∗∗∗ p<0.01; ∗∗ p<0.05; ∗p<0.1. S anda d e o s a e clus e ed a he p o ince le el. The coe icien
es ima es a e om he eg essions o he dummy a iable which akes a alue o 1 i he mo he egis e s
he child, and 0 o he wise on he gende no ms measu emen a iable (cho es and EmpGendRa io in wo
columns) and he con ol a iables
123
382 SERIEs (2024) 15:349–388
0
2500
5000
7500
Dec 18
Jan 19
Feb 19
Ma 19
Ap 19
May 19
Jun 19
Jul 19
Aug 19
Sep 19
Oc 19
No 19
Dec 19
Jan 20
Feb 20
Ma 20
Ap 20
May 20
Jun 20
Jul 20
Aug 20
Sep 20
Oc 20
No 20
Dec 20
Jan 21
Feb 21
Ma 21
Ap 21
May 21
Jun 21
Jul 21
egis a ion mon h
numbe o new egis a ions
Fig. 3 To al numbe o new egis a ions—mon hly. This igu e shows he e olu ion o he o al numbe o
new egis a ions o he Sma ick pla o m pe mon h, by people esiding in Spain
0
1000
2000
3000
4000
Dec 18
Jan 19
Feb 19
Ma 19
Ap 19
May 19
Jun 19
Jul 19
Aug 19
Sep 19
Oc 19
No 19
Dec 19
Jan 20
Feb 20
Ma 20
Ap 20
May 20
Jun 20
Jul 20
Aug 20
Sep 20
Oc 20
No 20
Dec 20
Jan 21
Feb 21
Ma 21
Ap 21
May 21
Jun 21
Jul 21
egis a ion mon h
numbe o new egis a ions
boys
gi ls
Fig. 4 To al numbe o new egis a ions by child gende —mon hly. This igu e shows, by child’s gende ,
he e olu ion o he o al numbe o new egis a ions o he Sma ick pla o m pe mon h, by people esiding
in Spain
123
SERIEs (2024) 15:349–388 383
0
2000
4000
6000
Dec 18
Jan 19
Feb 19
Ma 19
Ap 19
May 19
Jun 19
Jul 19
Aug 19
Sep 19
Oc 19
No 19
Dec 19
Jan 20
Feb 20
Ma 20
Ap 20
May 20
Jun 20
Jul 20
Aug 20
Sep 20
Oc 20
No 20
Dec 20
Jan 21
Feb 21
Ma 21
Ap 21
May 21
Jun 21
Jul 21
egis a ion mon h
numbe o new egis a ions
a he s
mo he s
Fig. 5 To al numbe o new egis a ions by pa en gende —mon hly. This igu e shows, by pa en ’s gende ,
he e olu ion o he o al numbe o new egis a ions o he Sma ick pla o m pe mon h, by people esiding
in Spain
0.2
0.3
0.4
Dec 18
Jan 19
Feb 19
Ma 19
Ap 19
May 19
Jun 19
Jul 19
Aug 19
Sep 19
Oc 19
No 19
Dec 19
Jan 20
Feb 20
Ma 20
Ap 20
May 20
Jun 20
Jul 20
Aug 20
Sep 20
Oc 20
No 20
Dec 20
Jan 21
Feb 21
Ma 21
Ap 21
May 21
Jun 21
Jul 21
egis a ion mon h
new egis a ions
mo he −gi l
mo he −boy
a he −gi l
a he −boy
Fig. 6 Pe cen age o new egis a ions by gende pai s—mon hly. This igu e shows he e olu ion o he
pe cen ages o new egis a ions by gende pai s o child en and pa en s o he Sma ick pla o m pe mon h,
by people esiding in Spain
123
384 SERIEs (2024) 15:349–388
0.0
0.2
0.4
0.6
0.8
Dec 18
Jan 19
Feb 19
Ma 19
Ap 19
May 19
Jun 19
Jul 19
Aug 19
Sep 19
Oc 19
No 19
Dec 19
Jan 20
Feb 20
Ma 20
Ap 20
May 20
Jun 20
Jul 20
Aug 20
Sep 20
Oc 20
No 20
Dec 20
Jan 21
Feb 21
Ma 21
Ap 21
May 21
Jun 21
Jul 21
egis a ion mon h
new egis a ions
p e−school
p ima y
seconda y
p e−school: 4−5, p ima y:6−11, seconda y: 12−16
Fig. 7 Pe cen age o new egis a ions by age ca ego ies—mon hly. This igu e shows he e olu ion o he
pe cen ages o new egis a ions by age ca ego ies o child en o he Sma ick pla o m pe mon h, by people
esiding in Spain
0.00
0.25
0.50
0.75
Dec 18
Jan 19
Feb 19
Ma 19
Ap 19
May 19
Jun 19
Jul 19
Aug 19
Sep 19
Oc 19
No 19
Dec 19
Jan 20
Feb 20
Ma 20
Ap 20
May 20
Jun 20
Jul 20
Aug 20
Sep 20
Oc 20
No 20
Dec 20
Jan 21
Feb 21
Ma 21
Ap 21
May 21
Jun 21
Jul 21
egis a ion mon h
new egis a ions
<25 h
25 h−50 h
50 h−75 h
>75 h
Fig. 8 Pe cen age o new egis a ions by income ca ego ies—mon hly. This igu e shows he e olu ion o
he pe cen ages o new egis a ions by income ca ego ies o he Sma ick pla o m pe mon h, by people
esiding in Spain
123
SERIEs (2024) 15:349–388 385
pe sis ency
comple ion
sessions
accu acy
delayed c.
−0.1 0.0 0.1 0.2
mo he a he
Fig. 9 Gende gap—co id sample. The Figu e shows he es ima ed gende gaps o di e en ou comes in
mo he s and a he s ull co id samples sepa a ely, based on Equa ion 1. Co id sample includes he membe s
egis e ed be ween 9 Ma ch 2020 and 31 May 2020, which co esponds o he i s wa e o he Co id-19
ou b eak in Spain. Con ol a iables a e age o he child, p esence o a heal h condi ion, o al numbe
o con ac s, ype o he i s con ac , p o ince and con ac yea -mon h ixed e ec s. S anda d e o s
a e clus e ed a he p o ince le el. Thin and hick lines ep esen he 95% and 90% con idence in e als,
espec i ely
pe sis ency
comple ion
sessions
accu acy
delayed c.
−0.1 0.0 0.1 0.2
mo he a he
Fig. 10 Gende gap—co id sample (siblings). The Figu e shows he es ima ed gende gaps o di e en
ou comes in mo he s and a he s siblings co id samples sepa a ely, based on Equa ion 2. Co id sample
includes he membe s egis e ed be ween 9 Ma ch 2020 and 31 May 2020, which co esponds o he i s
wa e o he Co id-19 ou b eak in Spain. Con ol a iables a e age o he child, p esence o a heal h condi ion
and pa en ixed e ec s. S anda d e o s a e clus e ed a he p o ince le el. Thin and hick lines ep esen
he 95% and 90% con idence in e als, espec i ely
123
386 SERIEs (2024) 15:349–388
1
6
11
16
21
26
−0.10 −0.05 0.00 0.05 0.10
mo he a he
Fig. 11 Di e en numbe o session cu -o s o comple ion a iable. The Figu e shows he es ima ed
gende gaps in comple ion ou come o di e en numbe s o session cu -o s in mo he s’ and a he s’ ull
samples sepa a ely, based on Equa ion 1. Con ol a iables a e he age o he child, p esence o a heal h
condi ion, o al numbe o con ac s, ype o he i s con ac , p o ince, and con ac yea -mon h ixed
e ec s. S anda d e o s a e clus e ed a he p o ince le el. Thin and hick lines ep esen he 95% and 90%
con idence in e als, espec i ely
pe sis ency
comple ion
sessions
accu acy
delayed c.
−0.1 0.0 0.1 0.2
mo he a he
Fig. 12 Age ixed e ec s in he main speci ica ion— ull sample. The Figu e shows he es ima ed gende
gaps o di e en ou comes in mo he s’ and a he s’ ull samples sepa a ely, based on Equa ion 1,whe e he
eg ession includes age ixed e ec s ins ead o he con inuous age a iable. Con ol a iables a e he age
ixed e ec s, p esence o a heal h condi ion, o al numbe o con ac s, ype o he i s con ac , p o ince,
and con ac yea -mon h ixed e ec s. S anda d e o s a e clus e ed a he p o ince le el. Thin and hick
lines ep esen he 95 and 90% con idence in e als, espec i ely
123
SERIEs (2024) 15:349–388 387
pe sis ency
comple ion
sessions
accu acy
delayed c.
−0.1 0.0 0.1 0.2
mo he a he
Fig. 13 No maliza ion wi hin age— ull sample. The Figu e shows he es ima ed gende gaps o di e en
ou comes in mo he s’ and a he s’ ull samples sepa a ely, based on Equa ion 1, whe e he ou come a iables
a e no malized wi hin each age. Con ol a iables a e he age o he child, p esence o a heal h condi ion, o al
numbe o con ac s, ype o he i s con ac , p o ince, and con ac yea -mon h ixed e ec s. S anda d
e o s a e clus e ed a he p o ince le el. Thin and hick lines ep esen he 95% and 90% con idence
in e als, espec i ely
pe sis ency
comple ion
sessions
accu acy
delayed c.
−0.1 0.0 0.1 0.2
mo he a he
Fig. 14 No maliza ion wi hin age—siblings’ sample. The Figu e shows he es ima ed gende gaps o
di e en ou comes in mo he s’ and a he s’ siblings’ samples sepa a ely, based on Equa ion 2,whe e he
ou come a iables a e no malized wi hin each age. Con ol a iables a e he age o he child, he p esence
o a heal h condi ion, and pa en ixed e ec s. S anda d e o s a e clus e ed a he p o ince le el. Thin and
hick lines ep esen he 95% and 90% con idence in e als, espec i ely
Re e ences
Anaya L, I ibe i N, Rey-Biel P, Zama o G (2022) Unde s anding pe o mance in es aking: he ole o
ques ion di icul y o de . Econ Educ Re 90:102293
And ew A, Ca an S, Cos a Dias M, Fa quha son C, K a man L, K u iko a S, Se illa A (2020) Inequali ies
in child en’s expe iences o home lea ning du ing he COVID-19 lockdown in England. Fisc S ud
41(3):653–683
Bake M, Milligan K (2016) Boy-gi l di e ences in pa en al ime in es men s: e idence om h ee coun ies.
J Hum Cap 10(4):399–441
Bala P, Oos e een M, Webbink D (2018) Tes sco es, noncogni i e skills and economic g ow h. Econ
Educ Re 63:134–153
Benjamini Y, Hochbe g Y (1995) Con olling he alse disco e y a e: a p ac ical and powe ul app oach o
mul iple es ing. J Roy S a Soc Se B (Me hodol) 57(1):289–300
Che y R, F iedman JN, S epne M (2024) The economic impac s o COVID-19: e idence om a new
public da abase buil using p i a e sec o da a. Q J Econ 139(2):829–889
De San Román AG, De La Rica S (2016) Gende gaps in PISA es sco es: he impac o social no ms and
he mo he ’s ansmission o ole a i udes. Es udios de Economía Aplicada 34(1):79–108
123
388 SERIEs (2024) 15:349–388
Del Boca D, Ogge o N, P o e a P, Rossi M (2020) Women’s and men’s wo k, housewo k and childca e,
be o e and du ing COVID-19. Re Econ Household 18(4):1001–1017
DeMa s CE, Bashko BM, Socha AB (2013) The ole o gende in es - aking mo i a ion unde low-s akes
condi ions. Res P ac Assess 8:69–82
Duckwo h AL, Quinn PD, Lynam DR, Loebe R, S ou hame -Loebe M (2011) Role o es mo i a ion in
in elligence es ing. P oc Na l Acad Sci 108(19):7716–7720
Ge ek ZE, Ge ek D, Neumeie C (2020) Explaining he gende gaps in ma hema ics achie emen and
a i udes: he ole o socie al gende equali y. Econ Educ Re 76:101978
Golin M (2021) Gende gaps in home schooling ime. mimeo
Ikeda M, Yamaguchi S (2021) Online lea ning du ing school closu e due o COVID-19. Jpn Econ Re
72(3):471–507
Lundbe g S, Pabilonia SW, Wa d-Ba s J (2007) Time alloca ion o pa en s and in es men s in sons and
daugh e s. Unpublished pape
Mammen K (2011) Fa he s’ ime in es men s in child en: do sons ge mo e? J Popul Econ 24(3):839–871
Mon olio D, Tabe ne PA (2021) Gende di e ences unde es p essu e and hei impac on academic
pe o mance: a quasi-expe imen al design. J Econ Beha O g 191:1065–1090
Nollenbe ge N, Rod íguez-Planas N, Se illa A (2016) The ma h gende gap: he ole o cul u e. Am Econ
Re 106(5):257–61
OECD (2019) PISA 2018 esul s (Volume I): wha s uden s know and can do. PISA. OECD Publishing,
Pa is
P ice J (2008) Pa en -child quali y ime does bi h o de ma e ? J Hum Resou 43(1):240–265
Rod íguez-Planas N, Nollenbe ge N (2018) Le he gi ls lea n! I is no only abou ma h... i ’s abou gende
social no ms. Econ Educ Re 62:230–253
Segal C (2012) Wo king when no one is wa ching: mo i a ion, es sco es, and economic success. Manage
Sci 58(8):1438–1457
Publishe ’s No e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps
and ins i u ional a ilia ions.
123
